Strategy Performance: July, 2009
All of the results below have been independently-audited in real-time by at least one third-party. Visit MarketSci.com for links to third-party audits and to learn more about accessing our strategies via a managed account or subscription.
This was a strange, strange month, dominated by the type of slow-motion grind up that has become a rarity in today’s mean-reversion dominated market.
MS and Scotty responded defensively by spending a good bit of the month in the safety of cash and ending the month more or less flat. YK did very well, generating big returns early in the month and then reducing exposure as the market and the abnormal market filter continued to ratchet up (read more).
And RH completely and utterly fell apart.
As readers know, RH is a product of our Timer Seeds program and is the only strategy I talk about on this blog that I didn’t directly develop. Because of that, and the fact that I try not to impose my own thoughts on our Timer Seeds (lest all of them start to look the same), RH does not include the abnormal market filter.
RH took a strong short position early in the rally and held on to it till the bitter end, ending the month down -19.4% and -25.9% in the S&P 500 and Russell 2000 variations respectively, and breaching its previous maximum drawdown.
So where do we go from here?
That’s the multi-million dollar question…
Scott Daly of Comprehensive Capital (who offers managed accounts tracking our programs) has removed RH from his recommended program list altogether, and is only trading the program by-request, because it exceeded his expected risk/reward thresholds. That’s one answer.
I’m not licensed to give specific advice for an individual’s specific financial situation, so I can only share my own very personal answer for my own very personal financial situation.
I’m maintaining a position in RH (but reducing exposure).
I think RH has a place in my balanced and diversified portfolio. Despite having an incredibly difficult time in this very specific market, the program has a solid overall track record. RH’s underperformance (and YK’s outperformance) over the last five months is a mirror image of what we saw in the depths of the financial crises last year when RH soared and YK plummeted. The two programs have played off of each other well in times of stress. I didn’t give up on YK then, and I’m not giving up on RH now.
I could of course be painfully wrong. Again, this is my very personal solution and isn’t necessarily appropriate for anyone else.
Having said all of that, I’m going to be working with RH’s developer in August to talk about adding the abnormal market filter to RH…group think be damned.
* * *
The equity curve below shows our combined real-time monthly performance since inception vs the S&P 500, assuming an equal investment in each family’s flagship program: MarketSci ProFunds, RH S&P 500, YK (A and B), and Scotty.
I’m going to start showing this graph logarithmically rather than linearly-scaled (making us look less sexy, but I think giving a more accurate view of performance).
COMBINED REAL-TIME PERFORMANCE vs S&P 500
SUMMARY OF ALL PORTFOLIOS SINCE INCEPTION
SORTED FROM MOST TO LEAST AGGRESSIVE *
Notes: (1) results account for transaction costs, but not subscription or managed account fees, (2) YK(B) returns reflect performance after addition of “abnormal market filter” in late October, 2008 (read more), (3) table sorted from highest to lowest measured volatility and may not reflect future performance.
* * *
One of the ways I justify spending my time on this blog is it gives me an opportunity once a month to share these independently-audited trading results. I really hope you enjoy my ramblings each week, but developing these proprietary strategies is really what I do, and I invite you to find out more about accessing our strategies via a subscription or managed account.
Happy Trading,
ms
To stay up to date with what’s happening at the MarketSci Blog, we recommend subscribing to our RSS Feed or Email Feed.
Filed under: My Performance | 7 Comments
Have you thought about combining the systems using a Risk Parity Approach (RPA), whereby low volatility is more highly allocated to than the high volatility. Should help address the RH allocation question. The RPA is nicely covered in Pangora’s downloadable document entitled “Risk Parity Portfolios: Effecient Portfolios Through True Diversification”
- Carl
RE to Carl – I’ve been doing a lot of work in the last couple of months re: the most effective way to combine strategies. I’m proposing a much more sophisticated method than risk-parity, but I’m not quite ready to share just yet. Still a WIP. michael
Michael,
Please keep us (and Scott) informed about future developments with RH and any abnormal market filter you may implement. I’m among the slaughtered last month but fortunately had YK to balance it out a bit. It does seem as though RH still has potential though.
RH’s performance realizes a concern that I have had with such models. IMHO one should have a thorough understanding of the conditions that would cause the model to fail disastrously. If the developer of the model doesn’t have that understanding, then I wouldn’t have confidence in using it. Also IMHO the peak draw down experienced by RH (which was more like 30%) qualifies as “disastrous”.
Unfortunately I’m left feeling that the information on the conditions under which various models (not just RH, and not just yours) will fall down is not forthcoming or is not well-understood. (by “fall down” I don’t mean a normal monthly setback but rather a prolonged and significant draw down).
RE to Brian: I agree and I disagree.
I agree that developers should do everything to build the best models they know how that will stand up against all market conditions. I agree that this was a punch in the eye for RH.
I disagree that a bad period somehow invalidates RH’s work. RH’s track record from inception makes this clear – I’ll venture a guess that it will take the market at least 10 years to catch up with where RH stands today – it has been an exceptional model that stumbled.
I preach incessantly that investors should be diversified in their approach, not necessarily in the traditional multi-asset class sense, but in the multi-strategy sense, because ALL strategies (and by ALL I mean ALL) are one trade away from their own black eye. For anyone who took that advice (even just within the MarketSci family of strategies, forgetting about the infinite number of other alternatives), this was just a boring flat month.
The point of all of that is to say, yes developers should try to understand what will make the model fall down, but at the end of the day, it’s impossible to really know. And the only solution as an investor is to have the eggs in many baskets.
michael
Hi Michael,
totally agree on the point about diversification (and not in the ‘traditional’ way). I’m not saying all models should stand up in all market conditions. But rather that there should be an understanding of the assumptions that would, should they not hold, cause the model to fail. eg, if you were to do Monte Carlo simulation using the model, what inputs would cause the model to do badly? And/or, has the model been backtested for a long enough period to see, historically, what the max draw down would have been (and what were the conditions at the time)?
In RH’s case I don’t mean to say that it is invalidated. Just that the current market conditions are providing inputs that result in undesirable outputs :-) As an investor, I’d like to have an idea what inputs are going to result in such undesirable outputs.
RE to Brian: I can definitely get on board with everything you wrote.
I don’t think Monte Carlo is an appropriate method for figuring out worse case scenarios – I think MC actually UNDER estimates potential downside risk.
I think the best way to understand potential downside is (a) as you mentioned, a very long and thorough backtest, and (b) a bit of logic.
An example of “a bit of logic”, today the YK model is very influenced by short-term mean-reversion (I say today b/c it’s an adaptive model and looked entirely different the last decade…or last year for that matter). A bit of logic says that a market that keeps moving in a single direction, day-after-day will break that model. The original YK(A) didn’t respect that very simple bit of logic and rode the market right off the cliff in last year’s financial crises b/c of a ridiculously long string of huge down days. Regardless of what historical tests showed (or how unlikely that event was), a bit of logic would have said “geez, maybe I should have a fallback plan in place in case that black swan comes to roost”. That fallback is now YK(B)’s abnormal market filter.
Okay enough rambling from me…good discussion and you got some creative juices flowing for me…michael